Feature-transforms Dataset

Computes the transformed features.

dai.transform(
  model,
  training_frame,
  validation_frame = NULL,
  testing_frame = NULL,
  validation_split_fraction = NULL,
  seed = NULL,
  fold_column = NULL,
  return_df = TRUE,
  progress = getOption("dai.progress", TRUE)
)

Arguments

model	The model to use for transformation.
training_frame	DAIFrame which will be used for training.
validation_frame	DAIFrame which will be used for validation.
testing_frame	DAIFrame which will be used for testing.
validation_split_fraction	If not having validation dataset, split ratio for splitting training dataset.
seed	Random seed for splitting.
fold_column	Fold column used for splitting.
return_df	Whether to return the transformed datasets as a list of data.frames or file paths.
progress	Whether to display a progress bar.

Value

A list of transformed datasets or file paths (see return_df parameter).

Details

The following datasets will be available upon successful completion:

• Training dataset (not to be used for cross-validation)

• Validation dataset for parameter tuning

• Test dataset for final scoring. This option is available if a test dataset was used. These datasets can be either obtained directly as list of data.frames if (return_df=TRUE), or a list of file paths that can be downloaded using dai.download_file.

See also

dai.download_file

Examples

dai.connect(uri = 'http://127.0.0.1:12345', username = 'h2oai', password = 'h2oai')
iris_dai <- as.DAIFrame(iris, progress = FALSE)
iris_splits <- dai.split_dataset(iris_dai, 'train', 'test', 0.8, progress = FALSE)
model <- dai.train(training_frame = iris_splits$train,
                   target_col = 'Species',
                   is_classification = TRUE,
                   is_timeseries = FALSE,
                   time = 1, accuracy = 1, interpretability = 10,
                   progress = FALSE)
iris_trans <- dai.transform(model, training_frame = iris_splits$train,
                            testing_frame = iris_splits$test, validation_split_fraction = .2)
head(iris_trans$test)